Simin Liu, Umed Ajani, Claudia Chae, Charles Hennekens, Julie E. Buring, JoAnn E. Manson. Long-term β-Carotene Supplementation and Risk of Type 2 Diabetes MellitusA Randomized Controlled Trial. JAMA. 1999;282(11):1073–1075. doi:10.1001/jama.282.11.1073
Author Affiliations: Division of Preventive Medicine (Drs Liu, Ajani, Chae, Hennekens, Buring, and Manson) and Channing Laboratory (Dr Manson), Department of Medicine, Brigham and Women's Hospital and Harvard Medical School; Department of Ambulatory Care and Prevention (Drs Hennekens and Buring), Harvard Medical School; Departments of Epidemiology (Drs Liu, Hennekens, Buring, and Manson) and Nutrition (Dr Liu), Harvard School of Public Health; Cardiology Division, Massachusetts General Hospital (Dr Chae), Boston
Context Recent data suggest a protective role of carotenoids in the development
of type 2 diabetes mellitus (DM), possibly via an antioxidant effect, but
no randomized trial has directly assessed the efficacy of β-carotene
to prevent DM.
Objective To determine whether long-term β-carotene supplementation reduces
the risk of developing type 2 DM.
Design, Setting, and Participants A total of 22,071 healthy US male physicians aged 40 to 84 years in
a randomized, double-blind, placebo-controlled trial, from 1982 to 1995. More
than 99% of the participants had complete follow-up (median duration, 12 years).
Intervention Subjects were randomly assigned to receive β-carotene (50 mg on
alternate days) or placebo.
Main Outcome Measure Incidence of type 2 DM.
Results A total of 10,756 subjects were assigned to β-carotene and 10,712
to placebo. Incidence of type 2 DM did not differ between groups: 396 men
in the β-carotene group and 402 men in the placebo group developed type
2 DM (relative risk, 0.98; 95% confidence interval, 0.85-1.12). The lack of
association between β-carotene supplementation and incidence of type
2 DM persisted despite multivariate adjustment. There was no evidence of benefit
when the period of risk was subdivided into years of follow-up or increasing
duration of treatment.
Conclusion In this trial of apparently healthy men, supplementation with β-carotene
for an average of 12 years had no effect on the risk of subsequent type 2
Increased free radical activity and high lipid oxidation impair glucose
disposal in the peripheral tissues and exacerbate diabetic complications.1- 3 These observations suggest
a role for oxidative stress in the pathogenesis of type 2 diabetes mellitus
(DM). Because of its extended system of conjugated double bonds, β-carotene
can scavenge peroxyl radicals and exert strong antioxidant activity,4,5 suggesting a protective effect against
the development of type 2 DM.6,7
Indirect evidence supporting such a protective role for β-carotene
comes from several observational studies relating increased intake of vegetables
that are rich in carotenoids with a lower risk of type 2 DM.8- 10
In addition, a recent dietary trial indicates that among people with impaired
glucose tolerance, those assigned to a diet with more vegetables have a lower
incidence of type 2 DM.11 It is possible, however,
that the observed reduction in risk associated with vegetables rich in carotenoids
may be due not to their β-carotene content but rather to other nutrients
in these foods or to other dietary or lifestyle factors. Moreover, biases
due to selection of subjects, misclassification of exposure, and residual
confounding cannot be fully addressed in observational studies,12
and no previous randomized trial has directly assessed the efficacy of β-carotene
to prevent type 2 DM.
To address this question, we examined the effect of long-term supplementation
with β-carotene on the incidence of type 2 DM in the Physicians' Health
Study, a randomized 12-year trial of 22,071 US male physicians.
The Physicians' Health Study was a randomized, double-blind, placebo-controlled
trial in which a 2×2 factorial design was used to evaluate the roles
of aspirin and β-carotene in the primary prevention of cardiovascular
disease and cancer. The study's methods have been detailed previously.13,14 In brief, at study entry in 1982,
22,071 male physicians aged 40 to 84 years were assigned randomly to receive
aspirin alone, β-carotene alone (50 mg on alternate days), aspirin plus β-carotene,
or both placebos. All men were free of a known history of myocardial infarction,
stroke, or cancer. For the current analysis, we excluded 603 men who reported
a history of diabetes before randomization. Thus, the final sample in this
analysis was 21,468 men. Participants were mailed monthly packs that contained β-carotene
or its placebo. The β-carotene component of the trial was terminated
as scheduled on December 31, 1995. By the end of 1995, 99.2% of all participants
were still providing questionnaire information on morbidity, including diagnosis
of type 2 DM. Follow-up for mortality was 100% complete. In both the β-carotene
and placebo groups, 80% of participants were still taking the study pills,
with an average compliance of 97%. In a validation study,15
men assigned to β-carotene had significantly higher serum β-carotene
concentrations than those assigned to placebo (2.24 vs 0.56 µmol/L [120
vs 30 µg/dL]).
Because type 2 DM was not the primary end point of this trial, medical
records were not requested to confirm the self-reported cases of diabetes.
All incident cases of diabetes occurring during follow-up were classified
as type 2 DM. Because all participants were physicians, their degree of access
to medical care is high and fairly homogeneous, and their reporting of cases
of type 2 DM was expected to be of high validity.16
For the current analysis, each participant accumulated follow-up time
beginning at baseline and ending on the month of diagnosis of type 2 DM or
censoring (death due to causes other than type 2 DM or December 31, 1995,
whichever came first). We calculated incidence rates of type 2 DM by dividing
the number of incident cases by person-years of follow-up. Relative risk was
estimated by dividing the rate in the β-carotene group by the rate in
the placebo group. We used Cox proportional hazards models to estimate the
relative risk of developing type 2 DM adjusting for age, aspirin assignment,
body mass index (calculated as weight in kilograms divided by the square of
height in meters) (BMI), smoking status, alcohol intake, physical activity,
history of high cholesterol level or hypertension, and use of multivitamins.
There were no clinically or statistically significant differences between
the 2 groups in terms of age, BMI, physical activity, cigarette smoking, alcohol
consumption, or other variables (Table 1).
After an average of 12 years of treatment and follow-up, no significant
benefit of β-carotene on risk of type 2 DM was evident (Table 2). During the follow-up period, 396 incident cases of type
2 DM in the β-carotene and 402 cases in the placebo group were reported,
for relative risk of 0.98 (95% confidence interval, 0.85-1.12). When the period
of risk was subdivided by years of follow-up, no benefit was observed for
any time period or duration of treatment (Table 2). The lack of association between β-carotene supplementation
and occurrence of type 2 DM persisted in multivariate analysis adjusting for
age, aspirin assignment, BMI, smoking status, alcohol intake, physical activity,
cholesterol level, hypertension, and use of multivitamins.
In this large-scale randomized trial among apparently healthy men, we
detected no significant change in risk of type 2 DM associated with 12 years
of β-carotene supplementation. The large sample size and long duration
of the trial allowed for precise estimates with narrow 95% confidence intervals,
excluding even a small effect of β-carotene.
In theory, poor compliance with the assigned treatment or an inadequate
dosage of β-carotene could explain the null findings. However, this explanation
is unlikely. The rate of compliance with β-carotene treatment was 85%
after 5 years and 78% after 12 years. The dosage of β-carotene increased
serum β-carotene concentrations by approximately 4-fold.15
This intake is roughly equivalent in its effect on blood levels to about 2
carrots a day and is above the level of dietary β-carotene consumption
that is associated with benefit in observational studies.14
Underdiagnosis of type 2 DM may still be a concern because the study
population was not screened for glucose tolerance and the diagnosis was self-reported.
However, this is not a plausible explanation for our findings since all participants
are physicians, who would be expected to report medical diagnoses accurately.
A validation study of self-reported diabetes in the Nurses' Health Study indicated
a high rate of agreement with medical record review.16
Given the randomized trial design, surveillance bias according to treatment
assignment is unlikely. Moreover, the identical distribution of baseline characteristics
in the β-carotene and placebo groups offers evidence that unknown confounding
factors were distributed equally between the groups.
Three observational studies have examined the relation between plasma β-carotene
level and degree of glucose intolerance or risk of type 2 DM.7,17,18
In a cross-sectional study of 109 hemodialysis patients, risk of diabetes
was inversely related to plasma β-carotene concentration.17
In a nested case-control study of serum β-carotene and risk of type 2
DM, participants in the highest tertile of serum levels of β-carotene
had a 55% lower risk of developing type 2 DM, but this association was greatly
attenuated after controlling for cardiovascular risk factors.18
Finally, a significant inverse relation between serum β-carotene level
and degree of glucose intolerance was observed in a random sample of 1665
US adults aged 40 to 74 years.7 Compared with
participants with normal glucose tolerance, participants with impaired glucose
tolerance had 13% lower β-carotene levels, and persons with newly diagnosed
diabetes had levels about 20% lower (P=.004 for linear
trend). Plasma levels of other carotenoids, such as lycopene and cryptoxanthin,
also were inversely related to glucose intolerance.
In several prospective cohort studies, increased consumption of vegetables
was associated with reduced risk of type 2 DM.8- 10
In a randomized trial of 577 people with impaired glucose tolerance who were
followed up for 6 years in Da-Qing, China, those assigned to a diet with more
vegetables had a 24% lower incidence of type 2 DM than the control group.11 This study, however, did not directly assess the
efficacy of β-carotene in the prevention of type 2 DM. Although rich
in β-carotene, vegetables have numerous other carotenoids that could
affect risk of DM.
In conclusion, β-carotene supplementation had no effect on the
risk of type 2 DM in this randomized trial of 12 years' duration. The results
of our study, however, should not be interpreted as completely refuting the
findings of observational studies that increased intake of vegetables that
are rich in carotenoids and other antioxidants may decrease the risk of type
2 DM. Our trial could not exclude the possibility that some carotenoids or
other nutrients other than β-carotene are responsible for the observed
association. Other antioxidants such as vitamin E may play a role in the prevention
of type 2 DM, but their efficacy still needs to be evaluated in randomized